Open-source AI Security Operations Center: alert fusion, LLM-agent triage, MITRE ATT&CK investigation, and a replayable decision ledger for every agent step. Self-hostable, runs with no API keys, MIT licensed. Ships an MCP server for Claude, Cursor and Continue.
Python agentic-aiai-securityalert-triageblue-teamcybersecuritydetection-engineeringincident-responselanggraphllm-agentsmcpmitre-attackmodel-context-protocolpurple-teamsecurity-operationsself-hostedsiemsoarsocthreat-detectionthreat-intelligence
"""Wave 5 (W5.3/W5.2) - the field-extraction / transform DSL."""
from __future__ import annotations
import pytest
from app.services.pipeline_transforms import (
TransformError,
apply_transforms,
validate_transforms,
)
def test_rename_moves_and_removes_source():
out = apply_transforms({"src": "v"}, [{"op": "rename", "from": "src", "to": "dst"}])
assert out.event == {"dst": "v"}
def test_dotted_paths_nest():
out = apply_transforms(
{"user": "alice"},
[{"op": "rename", "from": "user", "to": "actor.name"}],
)
assert out.event["actor"]["name"] == "alice"
def test_set_default_only_when_missing():
out = apply_transforms(
{"sev": "high"},
[
{"op": "set_default", "field": "sev", "value": "low"},
{"op": "set_default", "field": "conf", "value": 50},
],
)
assert out.event["sev"] == "high"
assert out.event["conf"] == 50
def test_lowercase_and_coalesce():
out = apply_transforms(
{"a": None, "b": "HELLO"},
[
{"op": "coalesce", "fields": ["a", "b"], "to": "greeting"},
{"op": "lowercase", "field": "greeting"},
],
)
assert out.event["greeting"] == "hello"
def test_extract_named_groups():
out = apply_transforms(
{"msg": "user=bob action=login src=10.0.0.1"},
[{"op": "extract", "field": "msg", "template": "user=%{WORD:user} action=%{WORD:action} src=%{IP:src_ip}"}],
)
assert out.event["user"] == "bob"
assert out.event["src_ip"] == "10.0.0.1"
def test_bad_op_warns_but_does_not_drop_event():
# 'from' present but no 'to' → the op raises, is caught as a warning, and
# the rest of the event survives (fail-open per op, never drop the event).
out = apply_transforms({"keep": "me", "src": "v"}, [{"op": "rename", "from": "src"}])
assert out.event["keep"] == "me"
assert out.warnings
def test_validate_rejects_unknown_op():
with pytest.raises(TransformError):
validate_transforms([{"op": "rm -rf"}])
def test_validate_rejects_unknown_grok_token():
with pytest.raises(TransformError):
validate_transforms([{"op": "extract", "field": "m", "template": "x=%{NOPE:v}"}])
def test_validate_rejects_too_many_ops():
with pytest.raises(TransformError):
validate_transforms([{"op": "drop", "field": "x"}] * 200)
def test_extract_literals_are_escaped_not_regex():
# Regex metacharacters in the template's literal text are matched literally,
# never interpreted — so no user data becomes a regex operator (ReDoS-proof).
out = apply_transforms(
{"m": "a.b=hello"},
[{"op": "extract", "field": "m", "template": "a.b=%{WORD:v}"}],
)
assert out.event["v"] == "hello"
# The literal "a.b" must not match "axb" (the dot is escaped).
out2 = apply_transforms(
{"m": "axb=hello"},
[{"op": "extract", "field": "m", "template": "a.b=%{WORD:v}"}],
)
assert "v" not in out2.event
def test_extract_input_is_length_bounded():
out = apply_transforms(
{"m": "x" * 10000 + " user=zed"},
[{"op": "extract", "field": "m", "template": "user=%{WORD:user}"}],
)
assert "user" not in out.event # trailing token beyond the cap is dropped
def test_original_event_is_not_mutated():
original = {"src": "v"}
apply_transforms(original, [{"op": "rename", "from": "src", "to": "dst"}])
assert original == {"src": "v"}